LONDON, Feb 6 - The one-size-fits-all AI trade that swept markets after the November 2022 launch of ChatGPT is losing cohesion as investors confront accelerating capital spending, growing debt and questions about who will capture the returns from the technology. Rather than moving in lockstep, stocks are now being separated by role in the AI ecosystem - who builds the infrastructure, who supplies inputs and who faces disruption.
That initial surge, which reached from semiconductor manufacturers and software vendors to raw-materials suppliers and businesses vulnerable to AI-driven disruption, helped push equity and debt markets higher and drew bubble warnings from regulators and investors. Tech giants including Microsoft, Amazon, Alphabet and Meta have since outlined plans for hundreds of billions of dollars of spending on the technology, but recent market volatility suggests investors are beginning to penalise cost without visible, immediate payoff.
Summary of market shifts
The market pullback this week has crystallised several patterns in how the AI theme is evolving. Four distinct trends stand out: a widening performance gap between AI infrastructure providers and other AI-linked firms; fragmentation among the most valuable U.S. technology names; a concentrated rally in South Korea driven by memory chip exposure; and signs that investors are demanding clearer evidence of returns on elevated capex.
1) "Picks and shovels" hold up better than downstream software
The rout in software stocks this week has accentuated the performance gap between the so-called AI "picks and shovels" - the hardware and infrastructure suppliers powering data-centre build-outs - and firms further down the chain that had been expected to benefit from generative AI-enhanced products.
In the United States, ServiceNow and Salesforce declined 12% and 9% respectively over the week. In Europe, data and analytics groups RELX and London Stock Exchange Group fell 16.4% and 6.3%. Semiconductor and data-centre-exposed names have also lost value this week but to a notably smaller degree, extending a divergence that had been forming between enablers of AI and companies seen as potential casualties or slower-to-monetise adopters.
"This divergence is not a vote against AI. It is a signal that investors are differentiating between who enables AI and who may be disrupted by it," Charu Chanana, chief investment strategist at Saxo, wrote in a note. Barclays equity strategists described a similar pattern in Europe, calling the dispersion in the region’s AI trade "extreme".
2) The Magnificent Seven are no longer monolithic
The cluster of the largest U.S. technology companies - often grouped as the "Magnificent Seven" - is showing internal splits as investors shift focus from headline capital commitments to whether those investments will generate measurable returns.
Portfolio managers at Goldman Sachs Asset Management highlighted in January that divergent AI and cloud strategies were beginning to break apart the Magnificent Seven narrative. Market moves have underscored that divergence. Microsoft and Meta each reported higher capex, yet Microsoft shares fell 10.4% on January 29 while Meta rose 10% over the same period. Alphabet reported a large jump in capex on Thursday, which pushed its shares down as much as 8% intraday before they finished the session flat. Amazon’s shares dropped 8.5% on Friday after the company announced a more than 50% increase in this year’s capital expenditure.
Mark Hawtin, head of global equities at Liontrust, warned of widening outcomes for the group: "There’s going to be huge divergence ... as a group they could well be market underperformers this year." He added, "You need to see a clear cause and effect. If they’re spending the money, are they getting a return for it? The market is no longer tolerating spending for spending’s sake." Over the past week, the Roundhill Magnificent Seven exchange-traded fund fell 5%, versus a 2% drop in the S&P 500.
3) South Korea’s memory-led rally stands out
With the ultimate winners among AI adopters still uncertain, many investors are placing bets on chipmakers, particularly those linked to AI-driven demand for memory. That positioning has propelled South Korea to the forefront of performance this year.
The main KOSPI index is up 20.8% year-to-date, compared with a 0.5% decline in the S&P 500 and a 4% gain in Europe’s STOXX 600. "From Q3 onwards - but it’s really only captured people’s attention in the last month - it (the AI capex trade) has now shifted really heavily to memory, which is a Korea trade," said Gerry Fowler, head of European equity strategy and global derivatives strategy at UBS.
South Korean chipmakers Samsung Electronics and SK Hynix are up 32% and 29% respectively so far this year. Morningstar Direct data shows flows into U.S.-listed Korean equity funds rose 20% in January, making them among the most popular fund picks for the month.
4) Investors demanding clearer returns on rising capex
Across these trends, a common thread is investor intolerance for unchecked spending without transparent pathways to profit. The immediate market reaction to capex announcements - where some companies were punished while others were rewarded - reflects heightened scrutiny over the effectiveness of those investments.
That shift is influencing sectoral and regional performance: hardware and data-centre supply chains are being viewed as nearer-term beneficiaries, while software, data and analytics firms are being evaluated more critically on their ability to monetize AI enhancements.
Implications for markets and investors
The evolving AI narrative suggests that allocations tied to the theme will continue to be reworked. Investors appear to be moving from indiscriminate exposure to differentiated bets based on the specific role a company plays in the AI value chain, its capital intensity and the clarity of its return profile.
Key points
- Hardware and data-centre suppliers are outperforming many software and analytics firms as investors separate AI enablers from those expected to be disrupted.
- Large-cap U.S. technology names are diverging in market reaction to higher capex, with the market increasingly evaluating return on investment rather than headline spending.
- South Korea’s market, driven by memory chipmakers, has outperformed major indices year-to-date amid concentrated investor interest in memory demand for AI.
Risks and uncertainties
- The rapid rise in capital expenditure among major technology firms raises the risk that investors could penalise companies that fail to demonstrate near-term returns, affecting large-cap tech and cloud services.
- Higher debt loads tied to AI spending could increase financial strain if revenue uplift from AI proves slower than anticipated, impacting both equity and credit markets.
- Market concentration in regions or sectors - such as Korea’s memory-focused rally - can expose investors to sector-specific cycles and idiosyncratic risks.
As the AI investment story matures, market participants are re-evaluating exposures with sharper distinctions between enablers, beneficiaries and those that may face disruption. The next phase will likely be determined by which companies can convert heavy spending into measurable, sustainable returns.